Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/n1bv_15.c @ 140:59a8758c56b1

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add source for PortAudio stable v190600_20161030
author Chris Cannam <cannam@all-day-breakfast.com>
date Tue, 03 Jan 2017 13:44:07 +0000
parents 7867fa7e1b6b
children
rev   line source
cannam@127 1 /*
cannam@127 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
cannam@127 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
cannam@127 4 *
cannam@127 5 * This program is free software; you can redistribute it and/or modify
cannam@127 6 * it under the terms of the GNU General Public License as published by
cannam@127 7 * the Free Software Foundation; either version 2 of the License, or
cannam@127 8 * (at your option) any later version.
cannam@127 9 *
cannam@127 10 * This program is distributed in the hope that it will be useful,
cannam@127 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
cannam@127 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
cannam@127 13 * GNU General Public License for more details.
cannam@127 14 *
cannam@127 15 * You should have received a copy of the GNU General Public License
cannam@127 16 * along with this program; if not, write to the Free Software
cannam@127 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
cannam@127 18 *
cannam@127 19 */
cannam@127 20
cannam@127 21 /* This file was automatically generated --- DO NOT EDIT */
cannam@127 22 /* Generated on Sat Jul 30 16:39:19 EDT 2016 */
cannam@127 23
cannam@127 24 #include "codelet-dft.h"
cannam@127 25
cannam@127 26 #ifdef HAVE_FMA
cannam@127 27
cannam@127 28 /* Generated by: ../../../genfft/gen_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 15 -name n1bv_15 -include n1b.h */
cannam@127 29
cannam@127 30 /*
cannam@127 31 * This function contains 78 FP additions, 49 FP multiplications,
cannam@127 32 * (or, 36 additions, 7 multiplications, 42 fused multiply/add),
cannam@127 33 * 78 stack variables, 8 constants, and 30 memory accesses
cannam@127 34 */
cannam@127 35 #include "n1b.h"
cannam@127 36
cannam@127 37 static void n1bv_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
cannam@127 38 {
cannam@127 39 DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
cannam@127 40 DVK(KP910592997, +0.910592997310029334643087372129977886038870291);
cannam@127 41 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
cannam@127 42 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
cannam@127 43 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
cannam@127 44 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
cannam@127 45 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
cannam@127 46 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
cannam@127 47 {
cannam@127 48 INT i;
cannam@127 49 const R *xi;
cannam@127 50 R *xo;
cannam@127 51 xi = ii;
cannam@127 52 xo = io;
cannam@127 53 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(30, is), MAKE_VOLATILE_STRIDE(30, os)) {
cannam@127 54 V Tb, TH, Tw, TA, Th, T11, T5, Ti, T12, Ta, Tx, Te, Tq, T16, Tj;
cannam@127 55 V T1, T2, T3;
cannam@127 56 T1 = LD(&(xi[0]), ivs, &(xi[0]));
cannam@127 57 T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
cannam@127 58 T3 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
cannam@127 59 {
cannam@127 60 V T6, T7, T8, Tm, Tn, To;
cannam@127 61 T6 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
cannam@127 62 T7 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
cannam@127 63 T8 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
cannam@127 64 Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
cannam@127 65 Tn = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
cannam@127 66 To = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
cannam@127 67 {
cannam@127 68 V T4, Tc, T9, Td, Tp;
cannam@127 69 Tb = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
cannam@127 70 T4 = VADD(T2, T3);
cannam@127 71 TH = VSUB(T2, T3);
cannam@127 72 Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
cannam@127 73 Tw = VSUB(T7, T8);
cannam@127 74 T9 = VADD(T7, T8);
cannam@127 75 Td = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
cannam@127 76 Tp = VADD(Tn, To);
cannam@127 77 TA = VSUB(Tn, To);
cannam@127 78 Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
cannam@127 79 T11 = VADD(T1, T4);
cannam@127 80 T5 = VFNMS(LDK(KP500000000), T4, T1);
cannam@127 81 Ti = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
cannam@127 82 T12 = VADD(T6, T9);
cannam@127 83 Ta = VFNMS(LDK(KP500000000), T9, T6);
cannam@127 84 Tx = VSUB(Tc, Td);
cannam@127 85 Te = VADD(Tc, Td);
cannam@127 86 Tq = VFNMS(LDK(KP500000000), Tp, Tm);
cannam@127 87 T16 = VADD(Tm, Tp);
cannam@127 88 Tj = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
cannam@127 89 }
cannam@127 90 }
cannam@127 91 {
cannam@127 92 V TI, Ty, T13, Tf, Tz, Tk;
cannam@127 93 TI = VADD(Tw, Tx);
cannam@127 94 Ty = VSUB(Tw, Tx);
cannam@127 95 T13 = VADD(Tb, Te);
cannam@127 96 Tf = VFNMS(LDK(KP500000000), Te, Tb);
cannam@127 97 Tz = VSUB(Ti, Tj);
cannam@127 98 Tk = VADD(Ti, Tj);
cannam@127 99 {
cannam@127 100 V T1d, T14, Tg, TE, TJ, TB, T15, Tl;
cannam@127 101 T1d = VSUB(T12, T13);
cannam@127 102 T14 = VADD(T12, T13);
cannam@127 103 Tg = VADD(Ta, Tf);
cannam@127 104 TE = VSUB(Ta, Tf);
cannam@127 105 TJ = VADD(Tz, TA);
cannam@127 106 TB = VSUB(Tz, TA);
cannam@127 107 T15 = VADD(Th, Tk);
cannam@127 108 Tl = VFNMS(LDK(KP500000000), Tk, Th);
cannam@127 109 {
cannam@127 110 V TM, TK, TS, TC, T1c, T17, Tr, TF, TL, T10;
cannam@127 111 TM = VSUB(TI, TJ);
cannam@127 112 TK = VADD(TI, TJ);
cannam@127 113 TS = VFNMS(LDK(KP618033988), Ty, TB);
cannam@127 114 TC = VFMA(LDK(KP618033988), TB, Ty);
cannam@127 115 T1c = VSUB(T15, T16);
cannam@127 116 T17 = VADD(T15, T16);
cannam@127 117 Tr = VADD(Tl, Tq);
cannam@127 118 TF = VSUB(Tl, Tq);
cannam@127 119 TL = VFNMS(LDK(KP250000000), TK, TH);
cannam@127 120 T10 = VMUL(LDK(KP866025403), VADD(TH, TK));
cannam@127 121 {
cannam@127 122 V T1g, T1e, T1a, Tu, Ts, TU, TG, TV, TN, T19, T18, Tt, TZ;
cannam@127 123 T1g = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1c, T1d));
cannam@127 124 T1e = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1d, T1c));
cannam@127 125 T18 = VADD(T14, T17);
cannam@127 126 T1a = VSUB(T14, T17);
cannam@127 127 Tu = VSUB(Tg, Tr);
cannam@127 128 Ts = VADD(Tg, Tr);
cannam@127 129 TU = VFNMS(LDK(KP618033988), TE, TF);
cannam@127 130 TG = VFMA(LDK(KP618033988), TF, TE);
cannam@127 131 TV = VFNMS(LDK(KP559016994), TM, TL);
cannam@127 132 TN = VFMA(LDK(KP559016994), TM, TL);
cannam@127 133 ST(&(xo[0]), VADD(T11, T18), ovs, &(xo[0]));
cannam@127 134 T19 = VFNMS(LDK(KP250000000), T18, T11);
cannam@127 135 Tt = VFNMS(LDK(KP250000000), Ts, T5);
cannam@127 136 TZ = VADD(T5, Ts);
cannam@127 137 {
cannam@127 138 V TW, TY, TQ, TO, T1b, T1f, TR, Tv;
cannam@127 139 TW = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), TV, TU));
cannam@127 140 TY = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), TV, TU));
cannam@127 141 TQ = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), TN, TG));
cannam@127 142 TO = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), TN, TG));
cannam@127 143 T1b = VFNMS(LDK(KP559016994), T1a, T19);
cannam@127 144 T1f = VFMA(LDK(KP559016994), T1a, T19);
cannam@127 145 TR = VFNMS(LDK(KP559016994), Tu, Tt);
cannam@127 146 Tv = VFMA(LDK(KP559016994), Tu, Tt);
cannam@127 147 ST(&(xo[WS(os, 10)]), VFMAI(T10, TZ), ovs, &(xo[0]));
cannam@127 148 ST(&(xo[WS(os, 5)]), VFNMSI(T10, TZ), ovs, &(xo[WS(os, 1)]));
cannam@127 149 {
cannam@127 150 V TT, TX, TP, TD;
cannam@127 151 ST(&(xo[WS(os, 12)]), VFNMSI(T1e, T1b), ovs, &(xo[0]));
cannam@127 152 ST(&(xo[WS(os, 3)]), VFMAI(T1e, T1b), ovs, &(xo[WS(os, 1)]));
cannam@127 153 ST(&(xo[WS(os, 9)]), VFNMSI(T1g, T1f), ovs, &(xo[WS(os, 1)]));
cannam@127 154 ST(&(xo[WS(os, 6)]), VFMAI(T1g, T1f), ovs, &(xo[0]));
cannam@127 155 TT = VFNMS(LDK(KP823639103), TS, TR);
cannam@127 156 TX = VFMA(LDK(KP823639103), TS, TR);
cannam@127 157 TP = VFMA(LDK(KP823639103), TC, Tv);
cannam@127 158 TD = VFNMS(LDK(KP823639103), TC, Tv);
cannam@127 159 ST(&(xo[WS(os, 13)]), VFMAI(TW, TT), ovs, &(xo[WS(os, 1)]));
cannam@127 160 ST(&(xo[WS(os, 2)]), VFNMSI(TW, TT), ovs, &(xo[0]));
cannam@127 161 ST(&(xo[WS(os, 8)]), VFMAI(TY, TX), ovs, &(xo[0]));
cannam@127 162 ST(&(xo[WS(os, 7)]), VFNMSI(TY, TX), ovs, &(xo[WS(os, 1)]));
cannam@127 163 ST(&(xo[WS(os, 11)]), VFMAI(TQ, TP), ovs, &(xo[WS(os, 1)]));
cannam@127 164 ST(&(xo[WS(os, 4)]), VFNMSI(TQ, TP), ovs, &(xo[0]));
cannam@127 165 ST(&(xo[WS(os, 14)]), VFNMSI(TO, TD), ovs, &(xo[0]));
cannam@127 166 ST(&(xo[WS(os, 1)]), VFMAI(TO, TD), ovs, &(xo[WS(os, 1)]));
cannam@127 167 }
cannam@127 168 }
cannam@127 169 }
cannam@127 170 }
cannam@127 171 }
cannam@127 172 }
cannam@127 173 }
cannam@127 174 }
cannam@127 175 VLEAVE();
cannam@127 176 }
cannam@127 177
cannam@127 178 static const kdft_desc desc = { 15, XSIMD_STRING("n1bv_15"), {36, 7, 42, 0}, &GENUS, 0, 0, 0, 0 };
cannam@127 179
cannam@127 180 void XSIMD(codelet_n1bv_15) (planner *p) {
cannam@127 181 X(kdft_register) (p, n1bv_15, &desc);
cannam@127 182 }
cannam@127 183
cannam@127 184 #else /* HAVE_FMA */
cannam@127 185
cannam@127 186 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 15 -name n1bv_15 -include n1b.h */
cannam@127 187
cannam@127 188 /*
cannam@127 189 * This function contains 78 FP additions, 25 FP multiplications,
cannam@127 190 * (or, 64 additions, 11 multiplications, 14 fused multiply/add),
cannam@127 191 * 55 stack variables, 10 constants, and 30 memory accesses
cannam@127 192 */
cannam@127 193 #include "n1b.h"
cannam@127 194
cannam@127 195 static void n1bv_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
cannam@127 196 {
cannam@127 197 DVK(KP216506350, +0.216506350946109661690930792688234045867850657);
cannam@127 198 DVK(KP509036960, +0.509036960455127183450980863393907648510733164);
cannam@127 199 DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
cannam@127 200 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
cannam@127 201 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
cannam@127 202 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
cannam@127 203 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
cannam@127 204 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
cannam@127 205 DVK(KP484122918, +0.484122918275927110647408174972799951354115213);
cannam@127 206 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
cannam@127 207 {
cannam@127 208 INT i;
cannam@127 209 const R *xi;
cannam@127 210 R *xo;
cannam@127 211 xi = ii;
cannam@127 212 xo = io;
cannam@127 213 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(30, is), MAKE_VOLATILE_STRIDE(30, os)) {
cannam@127 214 V Ti, T11, TH, Ts, TL, TM, Tz, TC, TD, TI, T12, T13, T14, T15, T16;
cannam@127 215 V T17, Tf, Tj, TZ, T10;
cannam@127 216 {
cannam@127 217 V TF, Tg, Th, TG;
cannam@127 218 TF = LD(&(xi[0]), ivs, &(xi[0]));
cannam@127 219 Tg = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
cannam@127 220 Th = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
cannam@127 221 TG = VADD(Tg, Th);
cannam@127 222 Ti = VSUB(Tg, Th);
cannam@127 223 T11 = VADD(TF, TG);
cannam@127 224 TH = VFNMS(LDK(KP500000000), TG, TF);
cannam@127 225 }
cannam@127 226 {
cannam@127 227 V Tm, Tn, T3, To, Tw, Tx, Td, Ty, Tp, Tq, T6, Tr, Tt, Tu, Ta;
cannam@127 228 V Tv, T7, Te;
cannam@127 229 {
cannam@127 230 V T1, T2, Tb, Tc;
cannam@127 231 Tm = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
cannam@127 232 T1 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
cannam@127 233 T2 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
cannam@127 234 Tn = VADD(T1, T2);
cannam@127 235 T3 = VSUB(T1, T2);
cannam@127 236 To = VFNMS(LDK(KP500000000), Tn, Tm);
cannam@127 237 Tw = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
cannam@127 238 Tb = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
cannam@127 239 Tc = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
cannam@127 240 Tx = VADD(Tb, Tc);
cannam@127 241 Td = VSUB(Tb, Tc);
cannam@127 242 Ty = VFNMS(LDK(KP500000000), Tx, Tw);
cannam@127 243 }
cannam@127 244 {
cannam@127 245 V T4, T5, T8, T9;
cannam@127 246 Tp = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
cannam@127 247 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
cannam@127 248 T5 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
cannam@127 249 Tq = VADD(T4, T5);
cannam@127 250 T6 = VSUB(T4, T5);
cannam@127 251 Tr = VFNMS(LDK(KP500000000), Tq, Tp);
cannam@127 252 Tt = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
cannam@127 253 T8 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
cannam@127 254 T9 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
cannam@127 255 Tu = VADD(T8, T9);
cannam@127 256 Ta = VSUB(T8, T9);
cannam@127 257 Tv = VFNMS(LDK(KP500000000), Tu, Tt);
cannam@127 258 }
cannam@127 259 Ts = VSUB(To, Tr);
cannam@127 260 TL = VSUB(T3, T6);
cannam@127 261 TM = VSUB(Ta, Td);
cannam@127 262 Tz = VSUB(Tv, Ty);
cannam@127 263 TC = VADD(To, Tr);
cannam@127 264 TD = VADD(Tv, Ty);
cannam@127 265 TI = VADD(TC, TD);
cannam@127 266 T12 = VADD(Tm, Tn);
cannam@127 267 T13 = VADD(Tp, Tq);
cannam@127 268 T14 = VADD(T12, T13);
cannam@127 269 T15 = VADD(Tt, Tu);
cannam@127 270 T16 = VADD(Tw, Tx);
cannam@127 271 T17 = VADD(T15, T16);
cannam@127 272 T7 = VADD(T3, T6);
cannam@127 273 Te = VADD(Ta, Td);
cannam@127 274 Tf = VMUL(LDK(KP484122918), VSUB(T7, Te));
cannam@127 275 Tj = VADD(T7, Te);
cannam@127 276 }
cannam@127 277 TZ = VADD(TH, TI);
cannam@127 278 T10 = VBYI(VMUL(LDK(KP866025403), VADD(Ti, Tj)));
cannam@127 279 ST(&(xo[WS(os, 5)]), VSUB(TZ, T10), ovs, &(xo[WS(os, 1)]));
cannam@127 280 ST(&(xo[WS(os, 10)]), VADD(T10, TZ), ovs, &(xo[0]));
cannam@127 281 {
cannam@127 282 V T1a, T18, T19, T1e, T1f, T1c, T1d, T1g, T1b;
cannam@127 283 T1a = VMUL(LDK(KP559016994), VSUB(T14, T17));
cannam@127 284 T18 = VADD(T14, T17);
cannam@127 285 T19 = VFNMS(LDK(KP250000000), T18, T11);
cannam@127 286 T1c = VSUB(T12, T13);
cannam@127 287 T1d = VSUB(T15, T16);
cannam@127 288 T1e = VBYI(VFNMS(LDK(KP951056516), T1d, VMUL(LDK(KP587785252), T1c)));
cannam@127 289 T1f = VBYI(VFMA(LDK(KP951056516), T1c, VMUL(LDK(KP587785252), T1d)));
cannam@127 290 ST(&(xo[0]), VADD(T11, T18), ovs, &(xo[0]));
cannam@127 291 T1g = VADD(T1a, T19);
cannam@127 292 ST(&(xo[WS(os, 6)]), VADD(T1f, T1g), ovs, &(xo[0]));
cannam@127 293 ST(&(xo[WS(os, 9)]), VSUB(T1g, T1f), ovs, &(xo[WS(os, 1)]));
cannam@127 294 T1b = VSUB(T19, T1a);
cannam@127 295 ST(&(xo[WS(os, 3)]), VSUB(T1b, T1e), ovs, &(xo[WS(os, 1)]));
cannam@127 296 ST(&(xo[WS(os, 12)]), VADD(T1e, T1b), ovs, &(xo[0]));
cannam@127 297 }
cannam@127 298 {
cannam@127 299 V TA, TN, TU, TS, Tl, TR, TK, TV, Tk, TE, TJ;
cannam@127 300 TA = VFMA(LDK(KP951056516), Ts, VMUL(LDK(KP587785252), Tz));
cannam@127 301 TN = VFMA(LDK(KP823639103), TL, VMUL(LDK(KP509036960), TM));
cannam@127 302 TU = VFNMS(LDK(KP823639103), TM, VMUL(LDK(KP509036960), TL));
cannam@127 303 TS = VFNMS(LDK(KP951056516), Tz, VMUL(LDK(KP587785252), Ts));
cannam@127 304 Tk = VFNMS(LDK(KP216506350), Tj, VMUL(LDK(KP866025403), Ti));
cannam@127 305 Tl = VADD(Tf, Tk);
cannam@127 306 TR = VSUB(Tf, Tk);
cannam@127 307 TE = VMUL(LDK(KP559016994), VSUB(TC, TD));
cannam@127 308 TJ = VFNMS(LDK(KP250000000), TI, TH);
cannam@127 309 TK = VADD(TE, TJ);
cannam@127 310 TV = VSUB(TJ, TE);
cannam@127 311 {
cannam@127 312 V TB, TO, TX, TY;
cannam@127 313 TB = VBYI(VADD(Tl, TA));
cannam@127 314 TO = VSUB(TK, TN);
cannam@127 315 ST(&(xo[WS(os, 1)]), VADD(TB, TO), ovs, &(xo[WS(os, 1)]));
cannam@127 316 ST(&(xo[WS(os, 14)]), VSUB(TO, TB), ovs, &(xo[0]));
cannam@127 317 TX = VBYI(VSUB(TS, TR));
cannam@127 318 TY = VSUB(TV, TU);
cannam@127 319 ST(&(xo[WS(os, 7)]), VADD(TX, TY), ovs, &(xo[WS(os, 1)]));
cannam@127 320 ST(&(xo[WS(os, 8)]), VSUB(TY, TX), ovs, &(xo[0]));
cannam@127 321 }
cannam@127 322 {
cannam@127 323 V TP, TQ, TT, TW;
cannam@127 324 TP = VBYI(VSUB(Tl, TA));
cannam@127 325 TQ = VADD(TN, TK);
cannam@127 326 ST(&(xo[WS(os, 4)]), VADD(TP, TQ), ovs, &(xo[0]));
cannam@127 327 ST(&(xo[WS(os, 11)]), VSUB(TQ, TP), ovs, &(xo[WS(os, 1)]));
cannam@127 328 TT = VBYI(VADD(TR, TS));
cannam@127 329 TW = VADD(TU, TV);
cannam@127 330 ST(&(xo[WS(os, 2)]), VADD(TT, TW), ovs, &(xo[0]));
cannam@127 331 ST(&(xo[WS(os, 13)]), VSUB(TW, TT), ovs, &(xo[WS(os, 1)]));
cannam@127 332 }
cannam@127 333 }
cannam@127 334 }
cannam@127 335 }
cannam@127 336 VLEAVE();
cannam@127 337 }
cannam@127 338
cannam@127 339 static const kdft_desc desc = { 15, XSIMD_STRING("n1bv_15"), {64, 11, 14, 0}, &GENUS, 0, 0, 0, 0 };
cannam@127 340
cannam@127 341 void XSIMD(codelet_n1bv_15) (planner *p) {
cannam@127 342 X(kdft_register) (p, n1bv_15, &desc);
cannam@127 343 }
cannam@127 344
cannam@127 345 #endif /* HAVE_FMA */